The American states have pursued several different approaches to create jobs and foster economic development, including reducing the social cost of labor. I show that state labor costs are markedly lower in 1995 than in 1970, based on a factor analysis of wages and state regulations affecting unions and the cost of labor. Corporate low-wage strategies, international and interstate competition for business, and the growing weakness of labor unions are posited to account for this. But regression analysis shows that lower labor costs have had no significant impact on rates of job creation since 1980, have reduced unemployment only slightly, and have been less effective in the latter respect than either tax cuts or other economic development policies. Moreover, reducing labor costs can have adverse consequences, such as rising income inequality, slower growth in gross state product and productivity, and a less competitive position for states in the international economy.

Cross-sectional analyses of American state data testing models in which state size is a meaningful predictor face a nonobvious problem of limited observations. Quite simply, the limited number of large states, and especially the presence of the uniquely large state of California, provides few observations to anchor regression estimates. We explore this problem with close attention to Gray and Lowery's (1996a) energy, stability, area (ESA) model of interest system density, both replicating their results with new data and highlighting the utility and limitations of analyzing unusual cases in comparative state research. After replicating the ESA model, several diagnostics for analyzing outliers, leverage, and influence are examined. We show how supplemental data analyses can be used to assess the source, severity, and, sometimes, the solution of the problem.

Research on Asian American politics is hampered by data limitations. Asian Americans comprise a small proportion of the population, and few political candidates are of Asian descent. However, because the Asian American population is growing quickly, interest in the group's political behavior has grown. One source of data that can be exploited to understand Asian American political behavior is the state of Hawaii. Hawaii provides a natural experiment since the majority of its citizens are Asian American and Asian political candidates are commonplace. This study of Hawaiian politics focuses on Asian American campaign finance behavior. I find that as Asian Americans locate themselves in more multicultural settings, they become more politically strategic, less focused on national-origin groupings, and more inclined to embrace a pan-ethnic identity.

Proposition 227, passed by California voters in 1998, aimed to dismantle bilingual programs in public schools and to replace them with English-only programs. Bilingual education, a long-standing program in California, involved mostly Hispanic students of limited English skills who were taught initially in their native language, and then were gradually transitioned into English-only classes. Using individual-level data from one southern California school district, I find that in 1998, before Proposition 227, limited-English-proficient (LEP) students enrolled in bilingual classes had lower scores in reading than LEP students who were not enrolled in bilingual classes, and who were, in general, more proficient in English. In math, bilingual students had test scores as good as those of non-bilingual LEPs. But in 1999, after Proposition 227, the same set of bilingual students had reading and math scores that were no worse than those of non-bilingual LEPs. Proposition 227, which interrupted bilingual programs and emphasized English instruction, did not set bilingual LEP students back relative to non-bilingual LEPs, and it may have even benefited them.

Researchers commonly apply inferential statistical procedures to population data from the 50 U.S. states as if they were estimating population parameters from sample statistics. This method is incorrect because with population data there is no need to make inferences about quantities that are already known. Instead, authors should simply provide evidence that their specified model provides a good fit to the data. Summary measures of variance as well as the full engine of Bayesian statistics perform this function. This research note demonstrates why the current practice of making inferences from population data with the null hypothesis significance test is wrong, provides some specific examples of problems in the literature, and gives prescriptive advice about correctly assessing and conveying empirical model results.